From ccc722827895ea7d11dba845d00ec6ce5c7ee98f Mon Sep 17 00:00:00 2001
From: Erica Marigold <hi@devcomp.xyz>
Date: Tue, 31 Dec 2024 12:46:45 +0000
Subject: [PATCH] fix: avoid duplicate directory entries with trailing slash

Fixed a bug in the directory tree builder where for directories there
would be two entries, where one would have a trailing slash, and the
other wouldn't.

The directory without the trailing slash would get linked to its
children in the entries lookup table, while the one with the trailing
slash would exist as a "ghost directory" with no children.

The fix involved sorting the entries to first handle directories, and
linking to the existing parent directory entries for children.
---
 lib/init.luau | 88 +++++++++++++++++++++++++++++++--------------------
 1 file changed, 53 insertions(+), 35 deletions(-)

diff --git a/lib/init.luau b/lib/init.luau
index 61fcf24..481b5ae 100644
--- a/lib/init.luau
+++ b/lib/init.luau
@@ -165,10 +165,9 @@ function ZipReader.parseCentralDirectory(self: ZipReader): ()
 		local crc = buffer.readu32(self.data, pos + 16)
 		local size = buffer.readu32(self.data, pos + 24)
 		local offset = buffer.readu32(self.data, pos + 42)
+		local name = buffer.readstring(self.data, pos + 46, nameLength)
 
-		local nameBuffer = buffer.create(nameLength)
-		buffer.copy(nameBuffer, 0, self.data, pos + 46, nameLength)
-		local name = buffer.tostring(nameBuffer)
+		print("got name:", name)
 
 		local entry = ZipEntry.new(name, size, offset, timestamp, crc)
 		table.insert(self.entries, entry)
@@ -178,43 +177,62 @@ function ZipReader.parseCentralDirectory(self: ZipReader): ()
 end
 
 function ZipReader.buildDirectoryTree(self: ZipReader): ()
-	for _, entry in self.entries do
-		local parts = {}
-		-- Split entry path into individual components
-		-- e.g. "folder/subfolder/file.txt" -> {"folder", "subfolder", "file.txt"}
-		for part in string.gmatch(entry.name, "([^/]+)/?") do
-			table.insert(parts, part)
-		end
+    -- Sort entries to process directories first; I could either handle
+	-- directories and files in separate passes over the entries, or sort
+	-- the entries so I handled the directories first -- I decided to do
+	-- the latter
+    table.sort(self.entries, function(a, b)
+        if a.isDirectory ~= b.isDirectory then
+            return a.isDirectory
+        end
+        return a.name < b.name
+    end)
 
-		-- Start from root directory
-		local current = self.root
-		local path = ""
+    for _, entry in self.entries do
+        local parts = {}
+        -- Split entry path into individual components
+        -- e.g. "folder/subfolder/file.txt" -> {"folder", "subfolder", "file.txt"}
+        for part in string.gmatch(entry.name, "([^/]+)/?") do
+            table.insert(parts, part)
+        end
 
-		-- Process each path component
-		for i, part in parts do
-			path ..= part
-			if i < #parts then
-				-- Create missing directory entries for intermediate paths
-				if not self.directories[path] then
-					local dir = ZipEntry.new(path, 0, 0, entry.timestamp, 0)
-					dir.isDirectory = true
-					dir.parent = current
+        -- Start from root directory
+        local current = self.root
+        local path = ""
 
-					-- Track directory in both lookup table and parent's children
-					self.directories[path] = dir
-					table.insert(current.children, dir)
-				end
+        -- Process each path component
+        for i, part in parts do
+            path ..= part
 
-				-- Move deeper into the tree
-				current = self.directories[path]
-				continue
-			end
+            if i < #parts or entry.isDirectory then
+                -- Create missing directory entries for intermediate paths
+                if not self.directories[path] then
+                    if entry.isDirectory and i == #parts then
+                        -- Existing directory entry, reuse it
+                        self.directories[path] = entry
+					else
+						-- Create new directory entry for intermediate paths or undefined
+						-- parent directories in the ZIP
+                        local dir = ZipEntry.new(path .. "/", 0, 0, entry.timestamp, 0)
+                        dir.isDirectory = true
+                        dir.parent = current
+                        self.directories[path] = dir
+                    end
 
-			-- Link file entry to its parent directory
-			entry.parent = current
-			table.insert(current.children, entry)
-		end
-	end
+                    -- Track directory in both lookup table and parent's children
+                    table.insert(current.children, self.directories[path])
+                end
+
+                -- Move deeper into the tree
+                current = self.directories[path]
+                continue
+            end
+
+            -- Link file entry to its parent directory
+            entry.parent = current
+            table.insert(current.children, entry)
+        end
+    end
 end
 
 function ZipReader.findEntry(self: ZipReader, path: string): ZipEntry